The article concerns the problem of forming a portfolio of state orders for implementing works at any of the three levels – federal, regional, and municipal – under a limited budget. Situations in which some of the works to be included in the portfolio can generate profit in the course of their implementation are the subject of consideration in the article. An approach to the analysis of these situations, along with that to finding an optimal structure of the portfolio in these situations, is proposed. The underlying idea of the approach consists of a) finding the maximum number of equally important projects that can be included in the portfolio taking into account that the profit that can be generated by implementing some of these works can be invested in some other works from the portfolio, followed by b) partitioning the portfolio into a set of independent groups of works. Both finding the above-mentioned maximum number and the partitioning are found by solving some Boolean and linear programming problems, respectively. Here, within each group, all the projects included in the group are financedfrom both the initial budget and the profit generated by some of the projects from the group. When such a partitioning into groups of works is feasible both technologically and legally, the administration in charge can form a portfolio of state orders for implementing works in the form of the mentioned groups of works and auction each group of the projects as a separate lot, i.e., can arrange combinatorial auctions in placing orders included in the portfolio.